so I watch this show, "Highway on my Plate" starring a Rocky and Mayur two non-skinny men, one from the south one from the north, one vegetarian the other non-vegetarian, as they travel all over India eating at local places.
It's getting to an interesting point now, because they've done so much traveling, that they begin to see some patterns emerge from their experiences. In other words, they've generated enough of a sample to see a trend. This is basically an empirical observation therefore: they say that no matter where they go, the appearance of a single dhaba by itself is never observed. Instead, there are long stretches of emptiness, followed by a sudden clustering of dhabas, followed by emptiness, and so on.
Why does this happen? For one reason, this is essentially the Hotelling model operating. In the simplest version of the Hotelling model, imagine two people competing for the custom of a local town. Call these two people restaurant owners. (Incidentally, Hotelling is the name of the guy who came up with this.) Where should they locate their restaurant? To simplify the problem to its essentials, imagine that the possible locations can be only on a single street, in fact, imagine the whole town to be arranged on a single street. So it's now just a matter of choosing where to locate on a single line.
Using assumptions on the cost functions of operating a restaurant (linear), and on the distribution of town residents on the street (uniform), it is easy to derive that the optimum location given that each hotel owners location impacts the number of consumers the other gets is the center of town. The basic intuition can be seen by seeing why any other location will not be optimal: assume one guy (A) gets to the center of town, now where should the other (B) locate? If B goes to the left of center, he will get the custom of all those to the left of his location, A will get all those to the right of his location (which is the center); but B and A will have to compete for those customers that locate between the their individual points.
Why will A get all those to the right of center? with a uniform distribution, the costs of finding a restaurant for any consumer will increase with the distance travelled, thus the result that A gets all those to the right of center, while B gets all those to the left of where B locates.
This competition between B and A can be alleviated if B shifts a little to the right. Now, B gets more consumers, while A gets the same number as before, although he probably loses a bit from before because some consumers he had now go to B. If A moves leftward he will gain a bit. B will respond by moving to the right. Eventually they will settle into the same place. However, and this is important, now either A or B can gain more by moving to the center. Thus the only place from where deviations are non-profitable for both A and B is the exact center.
Therefore, dhabas tend to cluster, and what's more, they should cluster somewhere in the middle of the cities. The episode on Ranchi I was watching appears to confirm the second result, but of course that is just one observation.
However there is also probably another reason, in that reciprocity - you benefit from having another dhaba nearby, in case one of your cooks doesn't show up, then you can borrow one from a nearby dhaba, with the understanding that in the future you will lend one of yours - can be another reason to locate nearby. Or they can take advantage of co-location and share electrical generators. This is more subtle reasoning, because it is a dynamic argument as opposed to the Hotelling model above which is static.
I just watched Exit through the Gift Shop, a documentary on street art that asks similar questions to those asked by the people who liked Iron Maiden and Motorhead and Metallica, were convinced this music would be big, however it was the glam rock bands that got big, which became a bit of a problem. Because it appears to prove that most people seem to prefer the inferior product, or are unable to tell the rip-offs from the originals.
6 comments:
It could be a case of the overall location being some sort of focal point of sorts - say near a petrol pump or a lay-by, or alternatively because that's a certain distance in terms of time from the nearest city. For example, you could have a dhaba open about 6 hours driving distance from the nearest city, since that might typically be the longest amount of time a driver may want to go without a break. (Reminds me of a Hitchcock quote about how the length of a movie should be in proportion to the capacity of the human bladder).
A clsuter may also come up around the first mover who builds a reputation for good food, and then as demand rises, more dhabas come up to meet that demand. Alternatively, one dhaba may not be able to provide a sufficiently wide range of foods, which might cause their clientele to be unhappy. So a cluster of dhabas might keep repeat customers happy, which in turn would lead to further business. Also, it allows for easier collusion on prices.
There may also be a selection bias involved in the selection of locales, which makes this clustering effect seem stronger than it is. These clusters will be better known and will be the sort of place a TV show might go to since there's more likelihood of getting good material there than at a stand-alone shack off the highway.
the middle point would turn to be a focal I believe; the "longest break" idea is also a possibility but that would presume those who run dhabas and those who eat at them understand driving constraints.
first mover building reputation is sort of like a stackleberg game with reputation added on.
i'm not sure about the variety of cuisine though; the other observation our pals Rocky and Mayur make is that most dhaba food is the same all over the country.
selection bias also can be dealt with the fact that (a) Rocky and Mayur say they find this pattern repeating (to this you may reply yes but the places have been chosen by the channel because "these clusters are better known" which then brings up the question why would they be better known - either the reputation story or the fact that co-location is responsible for generating least cost) and (b) this is something most of us have observed; the better places to eat tend to have a number of dhabas together.
ah but the 'middle point' of what, and how would that be calculated?
For the 'longest distance without a break' idea, it's not necessary for dhaba owners to calculate driving constraints ex ante. It may well be that they open wherever, but drivers would drive until they get tired and stop at that last dhaba. Then demand would be higher at that dhaba, and that location would flourish. What say?
it is an idea worth examining, I believe, is the cluster a result of careful Nash equilibrium or simply some sort of optimum stopping rule?
to your point: the focal point and the Nash could coincide. how can it be that way?
well, if our brain is capable of understanding parabolic curves of a falling ball without being endowed with a mind like that of Newton, why not?
it is also fair to ask if the Hotelling model was intended to address dynamic issues. If the model only points to a certain way to figure out a location initially at t = 0 then dynamic assessments are perhaps better viewed as extensions, not criticisms of the model.
Post a Comment